First of all you're supposed to be doing this in IPv6 not IPv4. There, you've had your IPv4 safety scolding. Secondly you're supposed to be using DNS, like the rest of the network world has for the last 20 years. So this should be about naming conventions not addresses. There, you've had your hard-coded-addresses safety scolding. Now to your question.
If you've got more than a class c worth of gear you probably shouldn't have it all on one subnet, just for practicality reasons. I agree grouping devices together can be helpful. You should have some sort of expansion scheme built into your whole architecture. So e.g. save off addresses for more cameras. But note that you're not likely to add another 1000 cameras to that one VMS the first 17 were connected to. So you should look at the big picture and maybe do one vms plus cameras each on their own subnet.
Lastly of course never ever ever do this in a vacuum because that's a good way to get in trouble with the local IT folks. Using a legitimate private address range like 172.16.20.0/24 is fine, as long as you've made some sort of effort to confirm you're not using the same subnet range as oh say the HVAC system, or the phone network.
And no it is very not cool to be caught assigning IP addresses over a blueprint at the back of the foreman's pickup truck on the job site. If you're not running off a spreadsheet that's under document control you're probably doing it wrong.